Risks, Research Agenda, and the Semantic Economy

Principal risks

Semantic infrastructure can concentrate power while appearing neutral. Registries can exclude dissenting meanings. Identity credentials can become access gates. Contribution graphs can become surveillance. Scores can reward performance over wellbeing. AI handlers can reproduce cultural bias while hiding behind technical complexity. Future financial incentives can encourage registration spam and enclosure of shared language.

The architecture is only defensible if it makes these risks visible and builds recourse before scale.

Falsifiable questions

The research program should test claims that could fail:

• Does a resolvable expression reduce ambiguity for independent implementers?
• Can participants understand the difference between a protocol facet, an expression, and a score?
• Does workshop-completed creator eligibility improve accountability without excluding valuable contributors?
• Can HMAC receipts and minimized saves support audit needs without reconstructing raw input?
• Do contribution graphs improve attribution compared with conventional authorship records?
• Can governance correct a harmful expression quickly and transparently?
• Does a Word Wallet produce useful nonfinancial attribution before any value-flow mechanism exists?

Evaluation before expansion

Research should publish sampling limits, negative findings, calibration failures, and governance incidents. Independent replication matters more than internal confidence. Changes to a handler or protocol need version-specific evidence.

The semantic economy hypothesis

The long-term hypothesis is that meaning maintenance, attribution, and coordination can become recognized economic infrastructure. This does not mean that every word becomes property or every interaction becomes a transaction. A healthier semantic economy would preserve commons, acknowledge contribution, and reduce the distance between institutional records and lived value.

The Phase 1 system is a small test of the prerequisites: canonical specification, public resolution, accountable execution, consent, identity eligibility, and nonfinancial attribution. It should be judged first on whether those prerequisites work.

Version history

• v0.1: archived conceptual paper.
• v0.2: revised web edition; introduces registered-expression grammar, current four-repository architecture, the working Cortisol Checker boundary, WellbeingIdentity eligibility, and explicit privacy and nonfinancial constraints.

References and grounding

• NIST AI RMF 1.0
• OECD Guidelines on Measuring Subjective Well-being, 2025 update
• W3C DID Core
• Internet of Value public implementation